Origin 200 Memory test failing -
nilgiri - 05-09-2020
Hi,
I've just acquired an Origin 200 which I'm attempting to get going.
Unfortunately I don't know much about its background.
It came with 4 x 64MB DIMMS (013-1372-002).
Whenever I turn it on I get a memory test fail.
I've tried multiple combinations of the DIMMs in Bank 0 (Slots 1 & 2)
I've also tried multiple combinations of the DIMMS in Bank 1 (Slots 3 & 4)
Regardless of the combination, I get a memory test failure:
Code:
IP27 PROM SGI Version 6.150 built 11:59:56 AM Sep 29, 2003
using BaseIO nic
Testing/Initializing memory ...............
*** Memory data bits stuck at 1 (act: 0x000000000000007e, exp: 0x0000000000000000)
memtest failed: stuck at 1
RSLT memtest FAIL diag_rc = 43 stuck at 1
*** Memory data bits stuck at 0 (act: 0xfffffffffffffc7f, exp: 0xffffffffffffffff)
memtest failed: stuck at 0
RSLT memtest FAIL diag_rc = 43 stuck at 0
A: +----------------------------------+
A: | Memory problems found in bank 0 |
A: | on Node board in slot n1. |
A: +----------------------------------+
*** WARNING: Disabling memory bank 0 in module 1 slot 1
DONE
*** No useable RAM installed. Need working and enabled memory in bank 0 or 1
*** Add working and enabled memory present in bank 0 or 1 and reset the system
I've cleaned the slots contacts and DIMM contacts with isopropyl (although they were very clean to start), but still get the failures.
At this stage I'm assuming I either have multiple faulty DIMMS, or a faulty board.
Does anyone have any suggestions for bringing this machine to life?
What does it mean by "enabled memory"?
RE: Origin 200 Memory test failing -
ghost180sx - 05-14-2020
You can have installed memory, but because of errors with them the system has not enabled any.
It is possible that if the RAM is bad in Bank 0, it doesn't even bother seeing the ones in Bank 1.
This system supports installing RAM in banks of 2. You need at least 2 DIMMs to operate in Bank 0.
Try only two DIMMs in Bank 0 (Slot 1 & 2) only. Does it report a failed DIMM?
You probably just have one or two bad DIMMs causing you grief. You might want to try and obtain a pair of know-good DIMMs for cheap off someone just to see if you can get it to work again and rule out the board.
http://www.sgistuff.net/hardware/systems/origin200.html#memory
RE: Origin 200 Memory test failing -
jan-jaap - 05-15-2020
First of all: the system should be able to overcome a bad first bank of RAM. It needs memory in the 2nd bank of course. It looks like your system has both banks disabled, or it would have booted up with either bank 0 or bank 1 disabled.
Does the system halt or does it drop you in a POD prompt? I think it should be able to reach a POD prompt even without any RAM installed. Otherwise you can force it to boot into POD using the dip switches on the MSC.
Once you make it into POD mode, you do
Code:
go cac
clearallogs
initalllogs
flush
reset
It may take a couple of minutes to reach CAC mode from POD (the 'go cac' step).
What the above does is reset the NUMA topology and initiate hardware (re)discovery. If you've got a broken bank of RAM it will *still* disable it the next time it performs POST, but hopefully the good bank of RAM will allow the system to boot up.
RE: Origin 200 Memory test failing -
ghost180sx - 05-15-2020
Oh that's fascinating. These seem really different than Octane!
I found this when searching about POD and CAC:
https://archive.org/stream/Origin2k_Hardware_Guide/hardware_djvu.txt
I also noticed this: (search doc for) "Testing Memory with POD Procedures"
But it says that is more for testing for intermittent failures.
but I think you should follow jan-jaap's advice!